[Yunme]

What this PR does / why we need it?

Adds devcontainer development environments that reproduce the PR gate pipeline locally:

• .devcontainer/cpu — CPU unit tests (cpu-ut stage)
• .devcontainer/npu — A3 two-card device tests (a3-2 partition)
• .devcontainer/a2 — A2 910B one-card device tests (a2-1 partition)
• .devcontainer/activate_npu.sh + select_npu.py — free-card detection
• .github/workflows/devcontainer_ci_consistency.yaml — CI consistency check between gate and devcontainer

Does this PR introduce any user-facing change?

No.

How was this patch tested?

Local devcontainer bring-up; CI workflow verification pending merge to base branch (GitHub does not run workflow changes from a PR).

• vLLM main: vllm-project/vllm@84030bb

[gemini-code-assist]

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request enhances the development experience by providing standardized devcontainer configurations that mirror the project's CI gate pipeline. By automating environment setup and NPU device selection, developers can now reliably reproduce test failures and validate changes locally before pushing to the repository.

Highlights

• DevContainer Configuration: Added dedicated devcontainer environments for CPU, NPU A3, and NPU A2 to locally reproduce the CI gate pipeline.
• NPU Resource Management: Introduced activate_npu.sh and select_npu.py to dynamically detect and filter available NPU devices for development environments.
• CI Consistency: Added a new CI consistency workflow to ensure parity between the gate pipeline and the local devcontainer configurations.

New Features

🧠 You can now enable Memory (public preview) to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Ignored Files

• Ignored by pattern: .github/workflows/** (1)
  • .github/workflows/devcontainer_ci_consistency.yaml

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize the Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counterproductive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[github-actions]

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

• A PR should do only one thing, smaller PRs enable faster reviews.
• Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
• Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

[gemini-code-assist]

Code Review

Suggested PR Title:

[CI][Feature] Add Dev Container configurations for CPU and NPU (A2/A3) environments

Suggested PR Summary:

### What this PR does / why we need it?
This PR adds VS Code Dev Container configurations for CPU, NPU A2 (single card), and NPU A3 (dual card) environments to simplify local reproduction of CI gate tests. It also includes helper scripts (`select_npu.py` and `activate_npu.sh`) to automatically detect and assign free NPU cards.

Feedback on the PR:
- In `.devcontainer/select_npu.py`, `subprocess.run` should specify `check=True` to prevent silent failures when running `npu-smi info`.
- In both `.devcontainer/a2/postCreateCommand.sh` and `.devcontainer/npu/postCreateCommand.sh`, the `MAX_JOBS` environment variable is hardcoded (to 23 and 46), which can cause Out-Of-Memory (OOM) errors on machines with fewer resources. It is recommended to dynamically set this using `nproc`.

### Does this PR introduce _any_ user-facing change?
No.

### How was this patch tested?
The configurations replicate the environment and steps used in `.github/workflows/pr_test.yaml`.

[gemini-code-assist]

The subprocess.run call does not specify check=True. If npu-smi info fails with a non-zero exit code (e.g., due to driver initialization issues or permission errors), subprocess.run will not raise an exception. Instead, it will return a CompletedProcess object with an empty stdout, causing the script to silently conclude that there are no free NPU cards available instead of reporting the actual error. Adding check=True ensures that any non-zero exit code raises a CalledProcessError, which will be properly caught and reported by the except block.

Suggested change

	out = subprocess.run(
	["npu-smi", "info"], capture_output=True, text=True, timeout=30,
	).stdout
	res = subprocess.run(
	["npu-smi", "info"], capture_output=True, text=True, timeout=30, check=True
	)
	out = res.stdout

[gemini-code-assist]

Hardcoding MAX_JOBS=23 assumes the host machine has at least 24 cores and sufficient memory to support 23 parallel compilation jobs. On developer machines with fewer cores or limited RAM, this can easily trigger Out-Of-Memory (OOM) crashes during kernel compilation. It is highly recommended to dynamically determine the number of jobs based on the available CPU cores, while allowing an override via environment variables.

Suggested change

	export MAX_JOBS=23
	export MAX_JOBS=${MAX_JOBS:-$(nproc)}

[gemini-code-assist]

Hardcoding MAX_JOBS=46 assumes the host machine has at least 48 cores and sufficient memory to support 46 parallel compilation jobs. On developer machines with fewer cores or limited RAM, this can easily trigger Out-Of-Memory (OOM) crashes during kernel compilation. It is highly recommended to dynamically determine the number of jobs based on the available CPU cores, while allowing an override via environment variables.

Suggested change

	export MAX_JOBS=46
	export MAX_JOBS=${MAX_JOBS:-$(nproc)}